Matching Hourly and Peak Demand by Combining Different Renewable Energy Sources A case study for California in 2020 --- Graeme R.G. Hoste Michael J. Dvorak Mark Z. Jacobson Stanford University Department of Civil and Environmental Engineering Atmosphere/Energy Program [email protected] Abstract In 2002 the California legislature passed Senate Bill 1078, establishing the Renewables Portfolio Standard requiring 20 percent of the state’s electricity to come from renewable resources by 2010, with the additional goal of 33% by 2020 (California Senate, 2002; California Energy Commission [CEC], 2004). More recently, some legislative proposals have called for eliminating 80% of all carbon from energy to limit climate change to an ‘acceptable level’. At the passing of the 2002 California bill, qualifying renewables provided less than 10% of California’s energy supply (CEC, 2007). Several barriers slow the development of renewables; these include technological barriers, access to renewable resources, public perceptions, political pressure from interest groups, and cost, to name a few. This paper considers only one technological barrier to renewables: integration into the grid. Many renewable resources are intermittent or variable by nature—producing power inconsistently and somewhat unpredictably—while on the other end of the transmission line, consumers demand power variably but predictably throughout the day. The Independent System Operator (ISO) monitors this demand, turning on or off additional generation when necessary. As such, predictability of energy supply and demand is essential for grid management. For natural gas or hydroelectricity, supplies can be throttled relatively easily. But with a wind farm, power output cannot be ramped up on demand. In some cases, a single wind farm that is providing power steadily may see a drop in or complete loss of wind for a period. For this reason, grid operators generally pay less for energy provided from wind or solar power than from a conventional, predictable resource. Although wind, solar, tidal, and wave resources will always be intermittent when they are considered in isolation and at one location, several methods exist to reduce intermittency of delivered power. These include combining geographically disperse intermittent resources of the same type, using storage, and combining different renewables with complementary intermittencies (e.g., Kahn, 1979; Archer and Jacobson, 2003, 2007). This paper discusses the last method: integration of several independent resources. In the pages that follow, we demonstrate that the complementary intermittencies of wind and solar power in California, along with the flexibility of hydro, make it possible for a true portfolio of renewables to meet a significant portion of California’s electricity demand. In particular, we estimate mixes of renewable capacities required to supply 80% and 100% of California’s electricity and 2020 and show the feasibility of load-matching over the year with these resources. Additionally, we outline the tradeoffs between different renewable portfolios (i.e., wind-heavy or solar-heavy mixes). We conclude that combining at least four renewables, wind, solar, geothermal, and hydroelectric power in optimal proportions would allow California to meet up to 100% of its future hourly electric power demand assuming an expanded and improved transmission grid. 1. Electric Grid Basics Electricity is perhaps the only commodity consumed the instant it is produced. In an idealized case, when someone turns on lights or air conditioning (holding all other loads constant), a power plant somewhere must ramp up its output slightly to meet the increase in demand. Fortunately, California has a sufficiently large number of electricity users that small increases in loads in some places are well balanced by small decreases in loads elsewhere. But aggregate 2 statewide demand still changes significantly over the day and year, with July demand typically increasing by two thirds from early morning to afternoon, and by about the same amount from January afternoon to July afternoon. These trends are seen below in our estimates of 2020 California statewide electricity demand (Figure 1). These curves were extrapolated from 2006 aggregate demand data taken from Oasis, an online database managed by the California ISO (California Independent System Operator, 2007). Data were averaged over one month to give the average hourly demands for January, April, July, and October. By averaging over a month, unusual demands from unseasonably hot or cold days, holidays, brownouts, etc., were taken into account. To estimate 2020 demand from the 2006 data, a 1.4% annual increase was assumed, consistent with the CEC’s figure of 1.22 – 1.49% per year (CEC, 2001b). 55000 50000 45000 40000 35000 30000 25000 power, MW 20000 15000 10000 January April 5000 July October 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 time of day, PST Figure 1: Estimated average hour-by-hour California electricity demands for 2020. Projected from 2006 values assuming 1.4% annual growth. 2006 demand data from California ISO. 2. Methods As mentioned above, our analysis is performed for the average day in each season, with January, April, July, and October taken as representative months. The average hourly demand curves shown above make up one end of the equation; a well-packaged supply makes up the other. Throughout this report, we show that we can match this average demand in these months with the average supply (calculated from modeled wind speeds and real insolations, see Section 4) from four different renewable resources—geothermal, wind, solar, and hydro—as well as some conventional baseload generation. One question raised is whether it is appropriate to draw conclusions for hour-to-hour grid management on any given day from an analysis in which both supplies and demand are averaged over the entire month. Looking at each site individually, this method could pose problems. It is very unlikely that the wind speed at Altamont Pass every July day at 3pm will be exactly the average 3pm July wind speed, for example. But when looking at the total power generation from wind turbines distributed all over the state, it is much more likely that the aggregate power production, among all sites, at every hour will be very close to the average. As with the example of a single energy user’s increasing load being balanced by another’s decreasing load, below average wind speeds at some farms will be compensated for by above average wind speeds at others. As with demand, the large-scale variation in wind power output over the day will become more noticeable and important than small-scale variations in individual locations. With rooftop 3 PVs in millions of locations, the chance of simultaneous cloud cover affecting all systems is extremely small. While it would never make sense to use average wind speeds or insolations to predict real hourly outputs from a single wind farm or rooftop PV system, we assume for this analysis it is reasonable to use average aggregated statewide renewable supply to predict aggregate statewide electric power output. This assumption is reasonable so long as we can also assume that the transmission system will be upgraded sufficiently 3. A Renewable Package to Meet the 33% RPS Given that the ‘aggressive’ California RPS target is 33% by 2020, an initial step of our model was to estimate the renewable capacities that could be used to meet that goal. The 2007 capacities of California geothermal, wind, solar, and hydroelectric power were: • 1,870 MW of geothermal in two major developments: the Geysers (north of San Francisco) and various plants in Imperial Valley (CEC, 2002) • 700 MW of solar power, roughly half rooftop photovoltaics and half solar thermal (Quaschning and Muriel, 2001) • 2,421 MW of wind power in five large farms: 710 MW at Tehachapi, 619 MW at San Gorgonio Pass, 586 MW at Altamont Pass, 415 MW at the High Winds Energy Center, 16 MW at Pacheco Pass, and 75 MW in smaller, distributed farms (American Wind Energy Association, 2007) • 13,500 MW of dependable hydro capacity distributed over the state, with an annual output of 4,100 MW (30.4% capacity factor) (CEC, 2001a, 2008a) Due to growing environmental and social concerns, as well as the fact that large hydro facilities (over 30 MW) do not qualify as renewable under the RPS, we assume that no new hydro plants will be built in California by 2020 (California Senate, 2002). If the only increase in wind and solar capacities through 2020 is from projects already proposed to the California Energy Commission (which includes a 4,500 MW expansion of the Tehachapi wind farm, set to go online by 2010, 2,927 MW of solar thermal projects in Southern California, and 3,000 MW of rooftop PVs as proposed in the California Million Solar Roofs Initiative), we find that the current hydro and geothermal capacities are more than sufficient to meet the low RPS target of 20% by 2020 (California Wind Energy Association, 2007; CEC, 2008b; California Public Utilities Commission [CPUC], 2007a, b). To scale up to the 33% RPS target, we retain the estimates of geothermal, solar PV and solar thermal capacities from our low scenario and increase total wind capacity until the annual renewable energy generated is 33% of total demanded energy. This condition is met with a statewide capacity of 16,000 MW. Here, the wind capacity is distributed as follows: one new farm of 2,400 MW is added at Eureka (offshore),1 and the remaining capacity is spread proportionally over the existing farms (except Tehachapi—held constant). This amounts to 5,210 1 A 2007 paper by Dvorak et al suggests a very good wind resource just offshore of Eureka. This site has consistently high wind speeds over the year, and a reasonable potential of 2,400 MW, taking into account undersea topography and local transmission considerations (Dvorak et al, 2004).